Hey Guys,
I have an applications running on a server, which reaches out to a MS SQL database. Every now and then (like this morning at 7am) we find that this application stops working. I did some digging and I can plainly see via a netstat that there's a bunch of stuck zombied connections. Right now I have about 30 or so connections from that server to my sql database. The states on the SQL database are "close_wait" and the states on the application server are "fin_wait_2". Now I'm fairly familiar with the TCP teardown sequence, however a mix of documentation contradicts itself leading me to question where the problem actually exists.
I'm confused about whether the application server needs to do more at this point, with it's connections being in fin_wait_2, they should have timed out by now, so is this an issue at the application server?
The sql server in close_wait suggests to me that the application server has not closed the connection properly, but the application in fin_wait_2 suggests to me that the application server did it's part, and is now waiting for the sql server to close the connection fully.
Can anyone shed any light to this?
Thanks.